In recent years, a new phenomenon in China's economic development is the increasingly fierce competition for talent between regions and cities. Talent is a vital factor of production that can generate core regional competitive advantages. The supply of talent in a city depends on its population base and educational resources as well as the cross-regional talent mobility. Talent mobility will lead to a profound evolution of the regional pattern of China's economy. Lately, all cities in China have introduced plans to attract talent and corresponding talent policies for the required concentration of talent. However, formulating accurate and effective talent policies to guide the rational flow of talent has been challenging for policymakers. Therefore, analyzing factors influencing intercity talent mobility, or the willingness of talent to move and the factors influencing it, in terms of theory and policymaking practice is crucial. Through a questionnaire survey, this study empirically analyzes factors determining intercity talent mobility, thereby establishing a scientific policy for talent mobility, promoting the rational flow of talent, and enhancing the comprehensive competitiveness of cities.
1 Literature Review
Hicks
[1] proposed that the main factor of mobility is a wage differential (Mahroum
[2]; Mitsakis
[3]), which results in a talent flow according to gender (Landes
[4]; Gries
[5]). Based on Huckschel' traditional trade growth theory, this study hypothesizes that productivity differences among regions and flow cost affect talent flow (Fischer and Frenkel
[6]). Analysis of the psychological cost of mobility reveals that the cost increases as the distance between inflow and outflow increases. The distance between regions has a lower constraint on mobility with improvements in education (Schultz
[7]). Further, people who have acquired a higher education are more likely to endure long-distance mobility (Suval and Hamilton
[8]). Gibson and Mckenzie argued that talent is more likely to flow into places with high technology, a comfortable environment, sufficient project funding, and a higher pay
[9]. Accordingly, talent inflow is often happened in developed areas, whereas talent outflow is always from areas with lower levels of economic development (Liu
[10]). Zhou, et al.
[11] suggested that inter-regional differences in terms of wages, institutional construction, and public service facilities are the main determinants of a high-level talent flow.
With global integration, scholars have focused on the flow of talent among countries. Kannankutty, et al.
[12] observed that in 2003, 16% of scientists and engineers in the US were immigrants. The cause of immigration was time, birth area, demographic characteristics, education, and job opportunities in the US. Wildasin identified that talent flow is mutually affected by taxation, the labor market, and other policies
[13]. Furthermore, personal factors, family, and social relations affected transnational talent flow (Coleman
[14]). Gender differences also affected talent flow (Ackers
[15]). Kannankutty, et al.
[12] suggested that family reasons, educational opportunities, and job opportunities (Yasin and Jauhar
[16]) influenced the choice to migrate
[11]. Scholars from China have suggested that geographical environment, national employment policy, the social security system, economic development, wages, and welfare result in talent flow.
Urbanization is essentially the congregation of production factors in space, which focuses on talent flow and aggregation. The greater the urbanization, the stronger the capacity for population aggregation (Zhang
[17]; Chen
[18]; Jiang, et al.
[19]; Huang
[20]). Research on urban talent flow has increased. Moreover, income gap between areas and imbalance in regional development leads to talent outflow. Cities with more employment opportunities are more likely to aggregate talent, and perfect public facilities and a better cultural atmosphere encourage the frequent flow and choice of talent between cities (Yan, et al
[21]; Zhou and Du
[22]; Li and Zhang
[23]; Yigitcanlar
[24]; Fan and Zhang
[25]; Holmes
[26]). Scholars have investigated the flow of scientific, technological, and academic talent, indicating the influencing factors such as regional economic growth, education, income, welfare, scientific research environment, housing, and family life (Ji and Zhu
[27]; Ji and Zhang
[28]; Ma
[29]; Hu
[30]; He
[31]).
Some shortcomings in the existing literature are as follows: First, research on talent flow tends to be theoretical rather than empirical. In particular, numerous qualitative studies have been conducted on factors influencing the willingness for talent mobility, and less empirical research has been observed. Second, research has focused on the flow of specific groups such as scientific and technological or college talent, with less focus on talent flow in less specialized groups. Most scholars have explored the current situation and reasons for the flow of special groups from several perspectives, focusing on regional investigation and general theoretical research; however, these studies lack in-depth and systematic analysis. Third, the literature has mainly focused on talent flow in specific regions or provinces, such as Beijing-Tianjin-Hebei and the Yangtze River Delta, with less research on talent flow across the entire country. Micro-level research has focused on talent flow from the enterprise perspective, with less research on talent flow between cities. Accordingly, this study empirically analyzes factors influencing the flow of unspecialized talent (i.e., of universal significance) from the national urban level. Moreover, the study assesses the differences between personal characteristics and the willingness for intercity talent mobility.
2 Research Design
The decision making guiding talent flow across cities is highly personalized and often influenced by personal characteristics and subjective factors. Gender, age, educational background, marital status, urban area, children, and occupation affect talent flow across cities.
Gender equality has become deeply rooted. However, the pressures faced by men and women in today's society differ, along with the attitudes and world outlook. Furthermore, the gender-based wage difference is a key factor in talent flow, leading to differential talent flow by gender. Men are usually more willing to relocate than women and are more likely to find a job, whereas women have stable jobs and income. Thus, we hypothesize the following:
Hypothesis 1 Gender affects the willingness for intercity talent flow, with men more likely to relocate than women.
Different age groups have different requirements concerning their working environment. Young people may pursue novelty and challenge, whereas older people may pursue stability. Therefore, age may influence intercity talent flow. Dai
[32] suggested that young people are more active and enterprising and will actively look for jobs and units conducive to their career development. Furthermore, they will never lose their jobs and units conducive to human capital investment to obtain greater development opportunities. Young people who find that their current work does not support their development are more likely to leave. Wang and Sun argued that the decision to relocate differs by age
[33]. Young people choose to relocate mainly for career development and dissatisfaction with salary and welfare, whereas older employees generally do not choose to relocate unless they are developing a second occupation and have a better prospect elsewhere. Older people generally have been working in their current unit for a long time and have a certain degree of experience and social status; thus, their willingness to relocate is small. Furthermore, the elderly are about to step into the ranks of retirees, and their life focus is changing or has changed from work to leisure; therefore, their enthusiasm and the requirements for work mobility are low. The younger the talent, the higher their mobility. Thus, we hypothesize the following:
Hypothesis 2 The willingness for intercity talent flows is affected by age: The older the talent, the lower the willingness.
With the development of society, changes in science and technology, and continuous improvement of urban development in emerging economies, the social demand for highly educated talent is increasing. Human resources have become a value-added resource pursued by cities. Businesses often attract human capital through high treatment. Meanwhile, well-educated people invest more time and energy into self-cultivation, hoping to get higher recognition and rewards in their work. Schultz argued that these costs are greater as the distance between inflow and outflow increases
[7]. As education increases, the distance between regions is less restrictive on talent flow. Therefore, more educated individuals are more likely to accept long-distance flow. Due to their advantages, they make higher requirements for their working environment and salary, in addition to having strong liquidity. Thus, we hypothesize the following:
Hypothesis 3 The willingness for intercity talent flow is affected by differences in education: The willingness of highly educated people is higher than that of those with lower education levels.
Owing to the effect of family and children's education, married individuals have weaker mobility and high stability and may be more likely to achieve better development in their existing environment. Marchs and Manari
[34] observed that married people were less likely to leave a stable job, whereas turnover was relatively high for unmarried individuals. Unmarried individuals are mostly students who have just graduated and entered the labor market. The transition from school to social life takes time, and they may still be looking for an ideal, satisfying job. Thus, students are more likely to leave a job with good opportunities elsewhere. Because unmarried individuals are not affected by the needs of their families, they are more concerned about future personal development, thereby having strong liquidity. We thus hypothesize the following:
Hypothesis 4 The willingness for intercity talent flow is affected by marital status: The willingness of unmarried individuals is higher than that of married people.
Furthermore, occupation affects talent flow. Differences exist among different professional talents in terms of organizational belonging and job satisfaction. In addition, the income gap between units is large, and differences exist in personnel mobility. Zhou, et al.
[11] revealed that the wage differential, institutional construction, and public service facilities mainly influence high-level talent flow. Researchers in scientific research institutions have strong independence and autonomy, and their satisfaction with salary and work is higher than that in other professions. Therefore, we hypothesize the following:
Hypothesis 5 The willingness for intercity talent flow is affected by occupational differences: Researchers have a higher willingness than other professionals.
A comparative analysis of eastern, central, and western China indicates significant differences in economic development, income, industrial structure, economic opportunity, cost of living, education, and medical care in terms of attracting talent from different regions. Gibson and Mckenzie
[9] documented that talent flows to places with a high level of technology, comfortable environment, sufficient project funding, and high salary. The inflow area is usually developed, and the outflow area is usually underdeveloped
[10]. Zhou and Du
[22] argued that brain drain from the central and western regions results from the material treatment gap as well as the working and living environment caused by the imbalance in regional development. Large differences in working and living environments and income distribution between China's east and west have led to the flow of a large pool of scientific and technological talent to the east. We therefore hypothesize the following:
Hypothesis 6 The willingness for intercity talent flow is affected by a city's region: The willingness in the central and western regions is stronger.
3 Empirical Analysis and Results
3.1 Model Construction
The logit regression model is suitable for regression analysis where the explanatory variable is a categorical variable. It is an ideal model for analyzing micro-individual willingness and its influencing factors. In this study, the willingness to move is the explanatory variable, a qualitative dichotomous variable. This study used the logit regression model to analyze factors influencing the willingness for talent mobility. The model is constructed as follows:
where is the probability indicating willingness for talent mobility, and is the probability indicating non-willingness for talent mobility. is the explanatory variable indicating the main factors influencing the willingness for talent mobility. denotes the intercept term, indicating the set of all explanatory variables. is the natural logarithm of the event occurrence ratio when it is zero. denotes the partial regression coefficient. It indicates the explanatory variable on the degree of influence of the explanatory variables when other explanatory variables are held constant. Further, denotes the log-event ratio caused by a one unit increase in the explanatory variable. denotes the average amount of change in the logarithm. For ease of interpretation, the natural logarithm is taken on both sides of the logistic model as follows:
When is a positive value, , i.e., the incidence ratio increases accordingly for each unit value increase. The change in the incidence ratio is represented by .
3.2 Data Sources and Variable Selection
We used a self-designed questionnaire to investigate factors influencing intercity talent mobility. The survey was divided into three parts: Background variables (gender, age, educational background, occupation, title, position, marriage, spouse's educational background, children, and city); talent mobility, including past talent mobility, reasons for future talent mobility, types, willingness, target cities for talent mobility (divided into first-tier cities, second-tier cities, third-tier cities, and others), and regions (divided into eastern, central, and western); the most attractive cities and factors influencing the attraction of talent to cities. The survey was conducted in cities across China from August 2019 to January 2020 through a combination of online and offline random sampling. Accordingly, 5, 290 questionnaires were distributed. After collation, 4, 915 questionnaires were correctly understood and completed, with a valid return rate of 92.91%. The Cronbach's alpha coefficient of this questionnaire was 0.825. The value was greater than 0.7, indicating that the standard error of measurement of the questionnaire was small, with good stability and consistency. The questionnaire covered 32 provinces, autonomous regions, and municipalities directly under the central government, prefecture-level cities, and most county-level cities, totaling 553. The questionnaires covered 31.7%, 8.7%, and 59.6% in the eastern, central, and western cities, respectively. The large proportion of cities in the west and the small proportion of cities in the centre is based on the "flight of the peacock to the south-east" and the massive brain drain from the west. Furthermore, the survey respondents included party and government organs, enterprise workers, and university students, covering all walks of life.
The subject of the study, i.e., the dependent variable was the willingness for intercity talent flow, which was a binary variable equal to 0 if there was no willingness to move and 1 otherwise.
The independent variables included in the model are as follows:
1) Gender (Ge) and age (Age): Gender is a binary variable, taking the value 0 for male (Ge0) respondents and 1 for female (Ge1). Age was divided into five categories: 25 years and below (Age0), 2635 years (Age 1), 3645 years (Age 2), 4655 years, and 56 years and above (Age 3). Because this variable contains five categories, the reference group was 25 years and below.
2) Educational background (Edu). This variable was divided into six categories: Junior high school or below; senior high school (technical secondary school, vocational high school); junior college; undergraduate; master's; and doctor. Three dummy variables were constructed: Junior college and below (Edu0), undergraduate (Edu1), and master's degree and above (Edu2). Junior college and below was the reference group (Edu0).
3) Marital status (Ma) and children (Ch). Marital status is a binary variable equal to 1 for married (Ma0) and 0 for unmarried (Ma1); with married as the reference group. There were six categories for the spouse's highest education (SE). Three 0-1 dummy variables were constructed: Junior college and below (SE0), undergraduate (SE1), and master's degree and above (SE2). Junior college and below was the reference group. Whether the respondent had children was captured via a dummy value equal to 1 for yes and 0 for no; with no being the reference group.
4) The area where the city is located (Re). Four categories were used in the questionnaire: Eastern, central, western, and Xinjiang, with eastern as the reference group. In the model, two categories were constructed: With 0 representing central and western cities (Re0) and 1 representing eastern cities (Re1). Central and western cities was used as the reference group.
5) Occupational status (Occ). There were seven response options in the questionnaire. Seven corresponding dummy variables were constructed: University staff (Occ1), medical institution staff (Occ2), scientific research institution staff (Occ3), party and government personnel (Occ4), enterprise staff (Occ0), college students (Occ5), and others (Occ6). Enterprise staff was the reference group.
The statistical descriptions of the variables are presented in Table 1.
Table 1 Demographic characteristics of the sample |
| Variables | Explanation of variables | Count (n=4915) | Percentage (%) |
| Ge | Gender | | |
| Male | 2250 | 45.8 |
| Female | 2665 | 54.2 |
| Age | Age | | |
| Under 25 years old | 1055 | 21.5 |
| 2635 years old | 1450 | 29.5 |
| 3645 years old | 1600 | 32.6 |
| 4655 years old | 755 | 15.4 |
| Over 56 years old | 55 | 1.1 |
| Edu | Education | | |
| Junior high school and below | 75 | 1.5 |
| Senior high school (technical secondary school, vocational high school) | 305 | 6.2 |
| Junior college | 760 | 15.5 |
| Undergraduate | 2545 | 51.8 |
| Master | 975 | 19.8 |
| Doctor | 255 | 5.2 |
| Ma | Marital status | | |
| Yes | 3130 | 63.7 |
| No | 1785 | 36.3 |
| Occ | Occupation | | |
| University Staff | 270 | 5.5 |
| Medical staff | 165 | 3.4 |
| Occ3 | Staff of scientific research institutions | 245 | 5 |
| Occ4 | Party and government personnel | 1370 | 27.9 |
| Occ0* | Enterprise staff | 915 | 18.6 |
| Occ5 | College Students | 660 | 13.4 |
| Occ6 | Other | 1290 | 26.2 |
| Re | Region | | |
| Re1 | East | 1560 | 31.7 |
| Re0* | Central | 430 | 8.7 |
| Western | 2925 | 59.6 |
| Note: * indicates that the data in this group is the reference group. |
3.3 Factors Influencing the Willingness for Intercity Talent Flow
SPSS 16.0 was used to test a logistical model. A test of the overall significance of the model was significant. Table 2 presents the results. In Table 2, B represents the degree of influence of factors on the willingness to flow. The positive and negative signs represent positive or negative influences, respectively. Value (Exp(B)) denotes the influence of each variable on the dependent variable. If the variable's OR value is greater than 1, the variable is a positive factor: The larger the value, the stronger the willingness to flow. If the OR value is less than 1, the variable is an avoidance factor. The smaller the value, the lower the willingness to flow. The corresponding B or the OR value is statistically significant when .
Table 2 Logit model of all respondents' willingness to move across cities |
| Variables in the Equation |
| | | B | S.E. | Wald | Df | Sig | Exp(B) |
| Step | Age | 0.758 | 0.27 | 7.909 | 1 | 0.005 | 2.134 |
| | Age | 0.642 | 0.321 | 4.003 | 1 | 0.045 | 1.9 |
| | Age | 0.44 | 0.341 | 1.668 | 1 | 0.197 | 1.553 |
| | Edu | 0.166 | 0.201 | 0.678 | 1 | 0.41 | 1.18 |
| | Edu | 0.482 | 0.18 | 7.146 | 1 | 0.008 | 0.618 |
| | SE | 0.03 | 0.215 | 0.019 | 1 | 0.889 | 0.971 |
| | SE | 0.098 | 0.221 | 0.196 | 1 | 0.658 | 0.907 |
| | Re | 0.154 | 0.051 | 8.954 | 1 | 0.003 | 0.857 |
| | Ma | 0.252 | 1.155 | 0.048 | 1 | 0.827 | 0.777 |
| | Ch | 0.308 | 0.286 | 1.159 | 1 | 0.282 | 1.361 |
| | Ge | 0.416 | 0.136 | 9.336 | 1 | 0.002 | 1.516 |
| | Occ | 0.42 | 0.339 | 1.539 | 1 | 0.215 | 1.522 |
| | Occ | 0.61 | 0.399 | 2.333 | 1 | 0.127 | 1.841 |
| | Occ | 0.775 | 0.353 | 4.829 | 1 | 0.028 | 2.172 |
| | Occ | 0.073 | 0.206 | 0.124 | 1 | 0.725 | 0.93 |
| | Occ | 0.537 | 0.333 | 2.605 | 1 | 0.107 | 0.584 |
| | Occ | 0.075 | 0.201 | 0.142 | 1 | 0.707 | 1.078 |
| | Constant | 1.041 | 1.327 | 0.616 | 1 | 0.432 | 0.353 |
| a. Variable(s) entered during Step 1: Age: 2635 years old; Age: 3645 years old; Age: 46 years old; Edu: undergraduate; Edu: master degree or above; SE: undergraduate; SE: master degree or above; region; marriage; children; gender; Occ: university staff; Occ: medical institution staff; Occ: scientific research institution staff; Occ: party and government organ staff; Occ: college students; Occ: other. |
From Table 2, the coefficient of gender is 0.416, SIG=0.0020.05, implying that the OR of men's intercity mobility is 1.516 times that of women, which is significant. Compared with women, men are more willing to move, thereby verifying Hypothesis 1.
The coefficients for age cohorts 2635 and 3645 years were 0.758, SIG=0.0050.05 and 0.642, SIG=0.0450.05, respectively. That is, the OR of intercity talent mobility for these two age groups was 2.134 times and 1.9 times that of the group under 25 years, respectively, and significant. Compared with most college students or those who only worked, young and middle-aged people with working experience and strong ability were more willing to relocate. The age of 46 years and above was significantly higher than 0.10 (SIG=0.1970.10), so older age had no significant effect on willingness to flow, thereby supporting Hypothesis 2.
Having a college degree or below did not significantly affect willingness to flow 0.10 (SIG=0.41 0.10). The coefficient for master's degree or above was (SIG=0.0080.05), lower than that for having a bachelor's degree or a college degree or below. This contravenes the literature and may be because cities focus on talents with a master's degree or above, giving them higher remuneration and better development prospects to retain talent so that they are more satisfied with their living and working environment. Although the major cities generally indicate a trend of increasing the educational background of individuals, the willingness of policy objects to move across cities is not high. Most of the individuals with a bachelor's or junior college degree hope to find a favorable environment for personal development with the improvement in their education, thereby verifying Hypothesis 3.
The coefficient for marital status was negative and insignificant. This is because unmarried people have higher job satisfaction and career development than married people. Married people, especially those with children, hope to find more opportunities for promotion and a more favorable external environment for personal development and benefit acquisition, so their willingness to flow is high. This finding is also consistent with the national situation in China, where there is an idiom called Mencius' mother moved three times, reflecting the importance that Chinese people attach to the educational environment for their children. Therefore, married people may have a higher willingness to relocate to have better protection for their children's education and medical care. Xia and Lu documented that laborers choose to flow to a city to obtain higher wage levels and employment opportunities in that city as well as to enjoy public services such as basic education and health care services in that city
[35]. Given the insignificance of this coefficient, Hypothesis 4 is not supported. The coefficient of employees in scientific research institutions was 0.775 (SIG=0.028
0.05), implying that their cross-city mobility was 2.172 times more than that of employees in enterprises. The coefficients for party, government officials, and college students were negative, and the coefficients for university staff, medical staff, and others were positive but not significant. Therefore, Hypothesis 5 is verified. The coefficient of region was
(SIG=0.003
0.05); that is, the cross-city flow rate of talent in eastern cities was 0.857 times that of central and western cities, and the coefficient was negative. This indicates that talent in eastern cities does not flow across cities, confirming the trend of talent flow from the central and western regions to the east. Hypothesis 6 is thus verified.
4 Analysis of the Differences in Factors Influencing Mobility Intentions in Different Regions
4.1 Analysis of the Model
The chosen classification regions were 1 = West, 2 = Central, and 3 = East. The differences in the factors affecting these three regions were examined. The specific results are presented in Table 3. For the western region, two factors were Edu (Master's degree), SE (Master's degree for spouse), i.e., compared with college students, the sample with a master's degree in the Western region was more reluctant to move with their spouse with a master's degree or higher. For the central region, there were four factors , namely Edu (Master), Occ (Researcher), SE (Spouse with Bachelor's degree), SE (Spouse with Master's degree), i.e., the sample in the central region was more reluctant to move for master's degree with their spouses with a bachelor's degree and above compared with college students. The researchers in the central region were more willing to move compared with corporate employees. For the eastern region, there were nine factors , namely Ge, Age, Age, Age, Occ (university personnel), Occ (other), SE (spouse with bachelor's degree), SE (spouse with master's degree), and Ch (having children). Compared with women, men were more willing to move; compared with those under 25 years, those aged 26 to 35, 36 to 45, and over 45 years in the east were more willing to move; university personnel and others were less willing to move compared with corporate employees; compared with those with spouses with college degrees, those with spouses with bachelor's and master's degrees were less willing to move; compared with those without children, those with children were more likely to move than those without children.
Table 3 Logit model of respondents' willingness to move across cities by region |
| Variable name | Explanation of variables | Western Region | Central Region | Eastern Region |
| Ge | Male | | 0.007 | |
| | | (1.652) | (0.048) | (2.061) |
| Age | 2635 years old | 0.193 | | |
| | | (0.873) | (1.724) | (4.044) |
| Age | 3645 years old | 0.043 | 0.211 | |
| | | (0.160) | (0.681) | (3.801) |
| Age | Over 46 years old | 0.028 | 0.181 | |
| | | (0.1000) | (0.545) | (3.586) |
| Edu | Undergraduate | 0.186 | 0.188 | |
| | | (1.042) | (0.706) | (1.852) |
| Edu | Masters | | | |
| | | (3.996) | (4.274) | (1.810) |
| Occ | University personnel | | 0.386 | |
| | | (1.684) | (1.170) | (0.295) |
| Occ | Medical staff | 0.109 | | |
| | | (0.320) | (2.368) | (1.876) |
| Occ | Scientific Researchers | 0.361 | | |
| | | (1.174) | (0.432) | (1.757) |
| Occ | Party and government bodies | 0.023 | 0.140 | |
| | | (0.132) | (0.617) | (2.496) |
| Occ | University students | 0.131 | 0.372 | |
| | | (0.478) | (1.138) | (1.822) |
| Occ | Other | 0.139 | 0.075 | |
| | | (0.767) | (0.325) | (0.086) |
| Ma | Unmarried | | 0.401 | 0.430 |
| | | (2.557) | (1.131) | (1.503) |
| SE | Spouse undergraduate | | | |
| | | (4.106) | (3.215) | (2.516) |
| SE | Spouse Master's degree | | | |
| | | (3.303) | (1.637) | (4.313) |
| Ch | With children | 0.200 | 0.050 | |
| | | (0.786) | (0.168) | (2.065) |
| Const | Constants | 0.556 | 0.206 | 0.356 |
| | | (1.368) | (0.420) | (0.972) |
| Note: , , denote , and , respectively. |
Accordingly, factors influencing the willingness to relocate differed in different regional cities. For the central region, main factors influencing the willingness to move were education as well as the type of occupation. In the east, factors influencing the willingness to move were mainly gender, age, occupation, education, and family. These results reflect regional differences. Thus, what are the individual-level factors such as gender, age, education, and occupation type as well as family-level factors such as marital status, spouse's education, and children that influence the willingness to relocate in different cities in different regions? In the following, we analyze factors influencing the willingness to move from one region to another at both levels.
4.1.1 Individual Level
The results of the logistic regression analysis, as indicated in Table 3, reveal that individual-level factors such as gender, age, education level, and type of occupation are statistically different and correlated with the willingness to move between cities in different regions.
Gender affects the willingness to move between western and eastern cities. Males in eastern cities have a stronger willingness to move compared with females. Previous research findings are more similar to the situation in eastern cities
[36], which is related to the fact that men in eastern cities are relatively better educated and move to cities mainly in search of better pay and more satisfying employment opportunities.
Age strongly affects the willingness to move in the east. Compared with those under the age of 25 years, those aged 26 years and over are less willing to move in western and central cities, whereas those aged 26 years and over are more willing to move in eastern cities. In other words, younger people in western and central cities are more willing to move, whereas younger and older people in eastern cities are more willing to move. Notably, the length of stay in a city is related to age and gender
[37], and dense and interconnected cities in the east offer a relatively large choice of employment opportunities for all age groups.
Educational attainment strongly affects the willingness for talent mobility in western and central cities. Among western and central city talents, a strong sense of reluctance exists to move among individuals with a master's degree and above in western and central compared with college students. This implies that college students are more willing to relocate among western and central city talents. On the one hand, this is due to the introduction of talent policies in western and central cities to increase efforts to retain and make good use of local talents with high education. On the other hand, the overall level of education among mobile talents in western and central cities is relatively low, mostly in small towns and agricultural household registration, and the human capital formed by receiving a certain high level of education is essential to promote urban migrant workers' choice of permanent migration
[38].
The type of occupation strongly influences the willingness for talent mobility in central and eastern cities. Compared with employees of enterprises, researchers in the central region are more willing to move, whereas university personnel and other personnel in the eastern region are not willing to move strongly. This is due to the level of regional economic development, salary and income levels, the research environment, research investment, which encourage the mobility of scientific and technological talents
[27]. The unbalanced level of economic development between the central and eastern parts of the country leads to the disparity in material treatment and the distribution mechanism, causing talent flow and creating the phenomenon of unreasonable distribution of talents
[22]. In the eastern part of the country, talents from universities and other personnel stabilize due to higher income levels and more job opportunities.
4.1.2 Family Level
The results of the logistic regression analysis, as presented in Table 3, indicate statistical differences and correlations between family-level factors such as spouse's education and children and the willingness to move to different regions of the city. Marital status slightly affects the willingness to move from one region to another.
Spouse's education strongly affects the willingness to move in different regions of the city. Spouses with a master's degree or higher in different regions have a stronger non-willingness to move compared with those with a college degree. In the central region, spouses with a bachelor's degree are also reluctant to move. This is due to the regions' efforts to retain talents by increasing the security and welfare of high-level talents in terms of housing, medical care, and spouse employment, resulting in a higher degree of job satisfaction and effectiveness.
The child factor strongly affects the willingness to move in eastern cities. Compared with those without children, talent from the eastern regions has a stronger willingness to move due to the pressure of competition for quality education for their children. In our analysis, a correlation exists between the fact that the majority of talent mobility in the eastern region is intra-provincial and intercity mobility and that this short to medium distance mobility is to some extent related to the ability of children to receive a better education
[39].
4.2 Goodness-of-Fit Test of the Model
The principle of the goodness-of-fit test is to compare the difference between the actual and predicted values of the model: Accordingly, the smaller the difference between the predicted and actual values, the better the model fit. The original hypothesis is that no significant difference exists between the predicted and actual values of the model. If the -value is greater than 0.05, the model is considered to have fitted the data well through the Hosmer–Lemeshow test. If the -value is less than 0.05, the hypothesis is rejected, indicating a significant difference between the predicted and actual values and that the model is a poor fit. Table 4 presents the results of the goodness-of-fit test for the logit model of respondents' willingness to move across cities by region.
Table 4 Hosmer-Lemeshow test |
| | Chi-squared | Df | p-value |
| Western Region | 7.022 | 8 | 0.534 |
| Central Region | 6.725 | 8 | 0.567 |
| Eastern Region | 10.215 | 8 | 0.250 |
According to the test results in Table 4, by region, the -values for the western, central, and eastern regions are all greater than 0.05, and none of them reject the original hypothesis. This indicates no significant difference between the predicted and actual values of the logit model concerning respondents' willingness to move across cities by region. The model has a good fit.
5 Analysis of Conclusions and Policy Implications
Based on the questionnaire survey data from 553 cities in China, this study uses logistic regression to examine the key factors influencing talent mobility across regions and analyzes the differences in the influencing factors of willingness for talent mobility at the individual and family levels in different regions. The main findings of the study are as follows: Middle-aged and young people (aged 2645 years) with some work experience are more willing to move than most university students or those who have just worked. Those with a bachelor's degree or higher are relatively more willing to move. We observed that the willingness to have a master's degree or higher was lower than the willingness to have a bachelor's degree, which is contrary to the literature but may be due to the fact that cities offer higher pay and better development prospects for high-level talent, thereby reducing the probability of their mobility. The willingness to move is stronger among individuals in scientific research institutions than those in other occupations. The general trend is that talent from central and western cities flows to eastern cities, whereas talent from eastern cities tends not to move across cities. We observed large regional differences in the willingness of talent to move across regions at the individual and family levels, which are closely related to the locational characteristics of cities in different regions and the individual family characteristics of mobile talent.
The factors influencing talent mobility across cities are diverse and interact with each other. These factors are both internal and external, and the degree of influence varies from large to small. The above analysis indicates that in the current context of talent mobility, where the main objective is to seek a better environment for development, economic income is the most significant influencing factor. It influences the role of other factors (such as occupation) on the willingness to move. Among these influencing factors, the attractiveness of the city (mainly including the level of economic development and employment opportunities brought by the development), marital and family status (employment of spouse and schooling of children), and personal characteristics (including age, gender, education and occupation) significantly affect the willingness to move. Clearly, the above empirical analysis indicates that the relation between many of these influencing factors and mobility intentions is not a simple linear relationship but is due to the interaction between the influencing factors.
Based on the findings of this paper, the government should have the following policy recommendations to improve the willingness to retain talent and achieve a rational layout of talent in the city.
1) Cities, especially those in central and western China, should form a differentiated industrial pattern with staggered functions; integrate functional, industrial, talent, and innovation chains; improve industrial agglomeration and urban attractiveness; promote urban economic development; create more employment opportunities; and retain and employ young and middle-aged talents with work experience.
2) Cities should provide a good working and living environment for talents, especially by strengthening family development guarantees and welfare benefits, (including spouse employment and children's education) to create a good sense of belonging and satisfaction for talents. While retaining highly educated and high-level talents, the city should increase its efforts to attract fresh university graduates from the region; further relax the restrictions on the age and overeducation of talent intake; and further increase the concentration of talent in the city.
3) Financial investment in research institutes, education, and other sectors should be increased to provide financial security for the city to attract and use scientific talent. Meanwhile, emphasis should be placed on building a science and technology innovation system; guiding enterprises to enhance their sense of independent innovation; and establishing a sound incentive mechanism to improve the attractiveness and stability of scientific researchers.
4) Furthermore, efforts should be made to coordinate the development of regional city economies and talents as well as the mechanism of cooperation between cities for talents, guide the flow of talent resources to cities in the central and western regions, strengthen cooperation between cities in the central and western regions and developed regions, and form a development pattern of complementary advantages.
Due to data limitations, factors such as urban public service facilities, social security systems, economic and social environments are not included in this paper.
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
PDF(156 KB)
Table 1 Demographic characteristics of the sample
